In addition to the Asus card that Nvidia sent over for the RTX 4070 Ti launch, MSI also sent over their flagship RTX 4070 Ti Suprim X. Their Suprim lineup always goes way over the top on cooling performance and the X means that this is also an overclocked card. So today I’m going to check out the MSI RTX 4070 Ti Suprim X to see if MSI is still keeping the Suprim styling as a clean opulent model which focuses on better build quality and performance. Without a Founders Edition for the new RTX 4070 Ti that leaves a big opening for anyone like me who may not want the crazy “gamer” styling that most cards have and the styling and build quality that the Founders Editon cards have. Will this be the best option when shopping for an RTX 4070 Ti? I’m going to check out what sets it apart and then put it through our test suite to find out!

Product Name: MSI RTX 4070 Ti Suprim X

Review Sample Provided by: MSI

Written by: Wes Compton

Amazon Affiliate Link: HERE

 

What is the RTX 4070 Ti all about?

Like with the RTX 4080 and RTX 4090 the new RTX 4070 Ti is also based on the Ada Lovelace and with that has brought a few big upgrades like the new 4th generation tensor cores for AI processing and 3rd generation ray tracing cores. The 4070 Ti is based on the AD104 GPU which is different from the RTX 4080 and RTX 4090 which were on the AD102 and AD103 GPUs. Being Ada based the 4070 Ti is designed to work with DLSS 3 which includes frame generation which helps with huge performance improvements when running DLSS 3 even for CPU-limited situations and games. The AD104 GPU in the 4070 Ti has 35.8 billion transistors compared to 45.9 billion on the RTX 4080. But in the specifications below where I compare the 4070 Ti with the 4080 and last generation’s 3070 Ti you can see how big of a jump it is over the 17.4 billion that the 3070 Ti had. The new GPU has 5 GPCs (Graphics Processing Clusters) which each have 6 TPCs (Texture Processing Clusters) which each TPC has two SMs (Streaming Multiprocessors) making for 60 in total. For comparison, the 3070 Ti had 48 and the 4080 has 76. This translates to 7680 CUDA cores on the 4070 Ti, 240 of the new 4th-gen Tensor cores, and 60 of the new 3rd-gen ray tracing cores.

Nvidia has the RTX 4070 Ti boost clock turned up compared to both the 3070 Ti which ran at 1770 MHz and the RTX 4080 which runs at 2505 MHz, the 4070 Ti has a boost clock of 2610 MHz. the 12GB of GDDR6X memory is clocked lower than the RTX 4080 though running at 1325 MHz compared to 1400 MHz which is still faster than last gens 3070 Ti which had its 8GB of memory running at 1188 MHz. The one area where the RTX 4070 Ti drops in specs is its memory interface which is 192-bit whereas both the 3070 Ti and the 4080 have 256-bit memory interfaces. With that its total memory bandwidth is lower than either of the other cards at 504 GB/sec vs 608 GB/s on the 3070 Ti and 716 GB/sec for the RTX 4080. Like with the other Ada-based GPUs, the 4070 Ti does have a much larger L2 Cache compared to the last gen, the 3070 Ti had 4MB whereas the 4070 Ti has 49 MB. That is lower than the RTX 4080 which has 64MB, as expected. It is built on the same TSMC 4N NVIDIA Custom Process as the other 4000 Series cards and has the same 3 DisplayPort and 1 HDMI configuration as last gen and this gen. That is HDMI 2.1 and DisplayPort 1.4a and while the extra doesn’t matter on this range of card for comparison AMD did include DisplayPort 2.1 on their 7900 XT which is the competitor for the RTX 4070 Ti. For TGP Nvidia has the 4070 Ti listed at 285 watts which is 5 watts less than the 3070 Ti and because they don’t have a Founders Edition of this GPU at all the power connection will depend on the model with some having the new 12vHPWR PCIe Gen 5 cable and others with two 8-pin PCIe power plugs.

RTX 3070 Ti

RTX 4070 Ti

RTX 4080

Graphics Processing Clusters

4

5

7

Texture Processing Clusters

24

30

38

Streaming Multiprocessors

48

60

76

CUDA Cores

6144

7680

9728

Tensor Cores

192 (3rd Gen)

240 (4th Gen)

3040 (4th Gen)

RT Cores

48 (2nd Gen)

60 (3rd Gen)

76 (3rd Gen)

Texture Units

192

240

304

ROPs

96

80

112

Boost Clock

1770 MHz

2610 MHz

2505 MHz

Memory Clock

1188 MHz

1313 MHz

1400 MHz

Memory Data Rate

19 Gbps

21 Gbps

22.4 Gbps

L2 Cache Size

4MB

49 MB

64 MB

Total Video Memory

8 GB GDDR6X

12 GB GDDR6X

16 GB GDDR6X

Memory Interface

256-bit

192-bit

256-bit

Total Memory Bandwidth

608.3 GB/sec

504 GB/sec

716.8 GB/sec

Texture Rate (Bilinear)

330 Gigatexels/sec

626 Gigatexels/sec

761.5 GigaTexels/sec

Fabrication Process

Samsung 8 nm 8N NVIDIA Custom Process

TSMC 4N NVIDIA Custom Process

TSMC 4N NVIDIA Custom Process

Transistor Count

17.4 Billion

35.8 billion

45.9 billion

Connectors

3 x DisplayPort

1 x HDMI

3 x DisplayPort

1 x HDMI

3 x DisplayPort

1 x HDMI

Power Connectors

1x12 pin

(Dongle to 2x 8-Pins)

2x PCIe 8-pin cables OR 300w or greater PCIe Gen 5 cable

1x16-pin

(Dongle to 3x 8-Pins)

Minimum Power Supply

600 Watts

700 Watts

750 Watts

Total Graphics Power (TGP)

290 Watts

285 Watts

320 Watts

Maximum GPU Temperature

93° C

90° C

90° C

PCI Express Interface

Gen 4

Gen 4

Gen 4

Launch MSRP

$599

$799

$1199

 

Before getting into testing I did also run GPUz to double-check that our clock speeds match up with the specifications. The MSI RTX 4070 Ti Suprim X is an overclocked card which can be seen with its boost clock speed sitting at 2775 Mhz to the 2610 MHz that Nvidia has as their reference boost clock. This is also higher than the 2730 MHz of the Asus TUF 4070 Ti OC that I took a look at for the launch. While GPUz didn’t have full support for the 4070 Ti at the time of testing it did document the BIOS revision and testing was done with the 527.62 driver which was provided by Nvidia to the press to test before the launch.

image 36

 

 

Log in to comment

We have 1518 guests and no members online

supportus